sql - Count CLOB Duplicates in a large Oracle Table

Question

I have an Oracle database table LOG_MESSAGES with a CLOB column called MESSAGE. Some of the rows contain the same MESSAGE.

For each MESSAGE which has at least a duplicate, I'd like to know the number of duplicates. Quite a number of these CLOBs are huge (> 100 kB), so converting to VARCHAR2 is out of question. Since many traditional methods such as GROUP BY do not work with CLOB, could someone please enlighten me?

For information, the table is very large (around 1 TB). So an optimised method would very much be appreciated.

Thank you in advance for your help.

Answer 1

dbms_crypto.hash can accept a CLOB and compute a hash. You can then group by the hash. Of course, computing a hash on a large CLOB is going to be an expensive process in terms of CPU consumption. If you have a large number of rows, it may take quite some time. You may want to compute and store the hash in one step and do a GROUP BY in a separate step.